How to Detect a Novelty Check out these

bestselling titles from

ChessCafe.com:
Emil Joseph Diemer (1908-1990) clearly preferred practice over theory.
When he had worked out a critical variation of his Blackmar-Diemer
Gambit (1 d4 d5 2 e4 dxe4 3 Nc3 Nf6 4 f3; or short: BDG) in joint
analyses with Georg Studier, it was still possible that he would play
something else in his next encounter with the same position. But the times
are changing, and today BDG players want to know precisely how to
react in a dangerous situation. In the last decades many BDG books and
magazines have been published, and while Diemers own Blackmar
Gemeinde (1956) mainly presented games, these recent publications no
Over the longer rate the beauty of a game higher than its theoretical relevance.
Now and then a line suffers a theoretical blow, made known in a new
Horizons book, and then the BDG player has to do his best to find a novelty that
repairs the damage. This is true for any opening, but when a BDG player
Dealing with d4
Deviations
ignores such a gap, he risks a lot more than a Queens Indian player. A by John Cox
Stefan Bcker case study: how to mend a critical BDG variation.

Fighting the
Anti-Kings Indians
by Yelena Dembo

Play through and download

the games from
ChessCafe.com in the
DGT Game Viewer.
The Belgian GM Albric OKelly,
The Complete a Correspondence World Champion
DGT Product Line
1. Trial and Error
Top TNT, Vol. 2
Geniuses, who can find the right solution on the spot, are rare. To learn by
your mistakes is, in the long run, at least as successful. The model of the
nave club player, who bangs out his first vague idea, only to correct it a
few seconds later, is emulated by the software which analyzes a chess
position: the first suggestion comes instantly, soon to be replaced by
something better. Our opening theory was developed by trial and error.
Armand Edward Blackmar, the inventor of the Blackmar Gambit 1 d4
d5 2 e4 dxe4 3 f3, died in 1888. In the same year Oskar Cordel mentioned
3e5 in Fhrer durch die Schachtheorie as a good reply. Had Blackmar
ever faced 3e5 in a game, he would have recognized the strength of the
move. Certainly he would have chosen the more accurate 3 Nc3 Nf6 4 f3
in his next game.

One of the critical lines of the BDG is the sequence that Albric OKelly
showed Diemer in June of 1956: 1 d4 d5 2 e4 dxe4 3 Nc3 Nf6 4 f3
4c6!

Black intends 5 fxe4 e5!. To delay taking on f3 has another advantage:

the move order 4exf3 5 Nxf3 c6 gives White additional options, such as
6 Ne5. In his book [1] Diemer recommended the following, adorned with
his usual exclamation marks, as the most forcing line against OKellys
idea: 5 Nxe4! Nxe4! 6 fxe4 e5 7 Nf3! exd4 8 Bc4!!, e.g., 8Bb4+ 9 c3!!
dxc3 10 Bxf7+!! Ke7 11 Qb3 cxb2+ 12 Qxb4+ Kxf7 13 Ne5+ Ke6 14 Qc4
+!! Kxe5 15 Bxb2+ and White wins. Some readers will recognize the
motif; Diemer had borrowed an old trick from a Caro-Kann Defense: 1 e4
c6 2 d4 d5 3 f3!? (von Bardeleben) 3dxe4 4 fxe4 e5 5 Nf3 exd4 6 Bc4
Bb4+? 7 c3! etc. However, in Diemers modified version, where the two
knights b1 and g8 are already exchanged, the move 7Be6! might be
even better than in the Caro-Kann variation. This was the line that I
recommended against the BDG when corresponding with Volker Drke
and Volker Hergert in 1986. A later thematic correspondence tournament
tested the OKelly Defense; in [2] the participants finally agreed that 4
c6 was a good weapon against the BDG, and that White probably had
nothing better than either 5 Nxe4 = or 5 fxe4 e5 6 Nf3 exd4 7 Qxd4 =.
These equal lines (the second case may even be slightly worse) are not
exactly what a BDG player is hoping for.

Of course, 4c6 does not refute the BDG. Its objective value may not be
greater than 4exf3 5 Nxf3 g6 (preferred by Georg Studier) or 4exf3 5
Nxf3 e6 (Joe Gallaghers BDG-antidote in Beating the Anti-Kings
Indians, 1996). However, in the last diagram few players take the safe
road 5 Nxe4, a majority hopes to transpose to standard BDG positions.
While BDG players know that they have to be cautious against 3e5 or
3f5, they often underestimate the OKelly Variation. Several
continuations are inaccurate:

(a) 5 Be3 Bf5 6 fxe4 Nxe4 is a Vienna Defense with a passive Be3 (it
belongs to f4).

(b) 5 Bg5 exf3 6 Nxf3 Bg4! (Volker Hergert) is a Teichmann Defense,

where a Be3 would be more useful, to protect the pawn d4.

(c) 5 Bf4 exf3 6 Nxf3 Bg4, and as in line b White would prefer a Be3
instead of the Bf4; for example, 7 Be2 Bxf3 8 Bxf3 e6.

(d) 5 a3 (Lev Zilbermints) 5exf3 6 Nxf3 g6, a Bogoljubow Defense,

where a3 is a bit slow.

And there is, finally, 5 Bc4 exf3 6 Nxf3 Bf5!

Black may claim that he is playing an improved version of 4exf3 5
Nxf3 c6 (6 Bc4?!), the so-called Ziegler Defense (but Id still prefer to
name this situation after OKelly, whose 4c6 is the more precise move
order). The diagrammed position at least offers more chances than
Whites other alternatives listed above (a-d). In 1998, I compared the
various anti-BDG systems a second time, my article [5] again
recommended the OKelly Defense and preferred Black in the
diagrammed position. However, couldnt my analysis have been wrong?
A modern computer still uses the same trial and error approach as I did
in 1986 and 1998, but it does it a bit faster.

2. Check Forcing Variations

Some theory of 1998 [5], just to see why Whites standard attack fails.

1 d4 d5 2 e4 dxe4 3 Nc3 Nf6 4 f3 c6! 5 Bc4 exf3 6 Nxf3 Bf5! 7 0-0 e6 8

Ne5

White threatens to take on f5 or on f7. The tactical 8Bxc2?!, deflecting

Whites queen from the protection of d4, seems to be strong, but only at
first sight. After 9 Nxf7! Kxf7, I found a strong improvement just by
checking the most forcing variations: 10 Qxc2! Qxd4+ 11 Be3! White is
clearly better. In the first game with this idea, St. Burke Ian Gallagher,
4NCL 2007, after 11Qxc4 12 Rf4! Qxf4 (the queen was trapped: 12
Qa6 13 Ra4) 13 Bxf4 Bc5+ 14 Kh1 Nbd7 15 Ne4 Nxe4 16 Qxe4 Nf6 17
Qe5 Bb6 18 Rf1?! Rhe8 19 Bg5 Bd8, Black was able to consolidate and
even win the game (0-1, 57). However, 18 Bg5! Rhf8 19 Rf1 would have
been stronger (+/). And the alternative 11Qxe3+ 12 Kh1 Bd6 13
Rae1, where I saw good compensation for White back in 1998, has now
become a +/, thanks to some quiet moves from the computer: 13Qh6
14 g4! (instead of the loud 14 Bxe6+?!) resp. 13Qd4 14 Ne4!
(instead of my 14 Re4). Thus it can be fruitful to check forcing variations,
but one has to keep an eye on quiet moves, too!

However, in the diagrammed position Black simply plays 8Bg6!, and

against this solid defense no one has found a promising attack. Whites
practical results are depressing.

3. Rearrange Old Motives

A small modification of a known motif can make a big difference. When I

recently returned to the BDG topic, just to see whether there was
something new, the next idea was a major surprise for me.
1 d4 d5 2 e4 dxe4 3 Nc3 Nf6 4 f3 c6! 5 Bc4 exf3 6 Nxf3 Bf5! 7 0-0 e6 8
Ng5!?

The so-called Alchemy Variation, whose theory was developed and

published on the Internet by Emanuel Rajmund [7] and members of the
Chesspublishing.com forum (cf. [6] and older threads). The knights jump
to g5 looks fresh and new in contrast to Ne5, White may now be able to
sacrifice on e6 but includes also an old motif. Against 8Bxc2, White
can just play 9 Nxf7, returning to the old tactical line above (while 9 Qe2!
[7] is probably even stronger). 8g6 9 h3 h6 10 Nxf7 may also be
playable for White. Blacks most reliable reply is the same as in the 8 Ne5
line.

8Bg6! 9 Ne2

The immediate sacrifice 9 Bxe6!? fxe6 10 Nxe6 was proposed by

SWJediknight, who correctly stated that after the natural continuation
10Qd7 11 Bg5 Qxe6 12 Re1 White maintains an attack [6]. However,
10...Qd6! settles the situation:

(a) On 11 Qe2, both possible replies 11...Kf7 and 11...Kd7 look reliable.

(b) 11 Bg5 Nbd7! (11...Qxe6 transposes to 10...Qd7, and 11...Kf7 allows

12 Bxf6 Qxe6 13 Be7+! or 12...gxf6 13 Ne4!) and Black seems to be safe,
e.g., 12 Re1 (12 d5 cxd5 [or Kf7] 13 Nxd5 Kf7 etc.) 12...Kf7 13 Bf4 Qb4,
and I dont see a good continuation (14 Qe2 Kg8 15 a3 Qa5).

(c) 11 Re1 is relatively best, but still insufficient: 11...Kf7 12 Bf4 (12 Bg5
is line b) 12...Qd7 13 Nc7 Bd6 (Na6!?, but the text is simple and good) 14
Bxd6 Qxd6 15 Nxa8 Nbd7 followed by Rxa8. Black is better.

Peter Leisebeins text move tries to prepare an assault on the vulnerable

squares f7 or e6. For example, 9c5?! [7] 10 Be3 Nc6 11 Nf4 cxd4 12
Bf2 Bf5 13 Re1 Be7 14 Nxf7! with a lasting attack and sufficient
compensation for the piece. But Black has a sufficient defense.

9Bd6! 10 Nf4 Bxf4 11 Bxf4

110-0! /+

An improvement by OldGrizzly [6]. I had found a flaw in the main

line 11Nd5 12 Bc1 [7], namely 12 h4! when 12h6 13 Nxe6 fxe6 14
Qg4 Bf5 15 Qxg7 Rh7 16 Qg8+ Kd7 17 Qg3 Nxf4 18 Rxf4 Na6 19 Rxf5
exf5 20 Be6+ leads to some chances for White, resembling the endings in
a Muzio Gambit. However, by just changing the move order OldGrizzly
has pushed White back into the cold. There could follow 12 a4 Nd5 13
Bc1 h6 14 Nf3 Nd7 /+, and White has nothing at all to show for his
sacrificed pawn.

4. Call a Grandmaster

When nothing else helps, make a phone call and ask a grandmaster. In
this case it was Lev Gutman, who calmly listened to my latest dubious
attempt to rescue the line, 7 g4??! (more on this below). One hour later he
called back. No, he could not bring himself to believe in my raving tactics
with 7 g4, but he had found a more positional set-up.

1 d4 d5 2 e4 dxe4 3 Nc3 Nf6 4 f3 c6! 5 Bc4 exf3 6 Nxf3 Bf5! 7 Bg5 e6 8

Nh4!? Bg6 9 Nxg6 hxg6 10 Qd3!

Gutmans key idea is to castle long (in contrast to old games with Ne5-
xg6 plus 0-0), play h4 and Rh3 (to be able to take back on c3 with the
rook, if necessary) and Rh3-f3, to pin Blacks pieces to the weak pawn on
f7. The text move includes motifs of Bxe6 and Qxg6+, but in the long run
the queen may belong on e2, to exert pressure on the e-file and assist an
advance of the kingside pawns.

10Be7

Or 10Qa5 11 h4!; for example,

(a) 11Qf5 12 Qe3 Qxc2? 13 0-0 +/.

(b) 11Bb4 12 Bxe6 0-0 (12fxe6? 13 Qxg6+ Kf8 14 0-0 +) 13 0-0

fxe6 14 Bxf6 Nd7 15 Bg5 +=.

(c) 11Qxg5 12 hxg5 Rxh1+ 13 Ke2 Rxa1 14 gxf6 Nd7 15 Bxe6 Nxf6
16 Nd5! unclear.

(d) 11Nbd7 12 0-0-0 0-0-0 13 Qe2! Bb4 14 Rh3. So far analysis by

Gutman. White has sufficient play for the pawn, e.g., 14Nb6 15 Bb3
Rd7 16 a3 Bd6 17 Rf3 Nbd5 18 Na4!.

11 0-0-0 Nbd7

Or 11Nd5 12 Bxe7 Nxe7 13 Ne4, with compensation for White.

12 h4 a5

Or 12b5 13 Bb3 a5 14 a4 b4 15 Ne4, and Black still has to find shelter

for his king.

13 Bb3 Nb6 14 a4

I didnt find an example of the position after 10 Qd3 in various BDG

books or in a database of 7,000 BDG games. The reason for neglecting
the idea may be that it requires the vision of a grandmaster to recognize
the hidden defects in Blacks position. Where will the Ke8 go? He may be
safer on c8, but then the weakness f7 plays an even greater role and will
make it be almost impossible to activate that pawn. Whites pawns on the
kingside with their urge to advance are not a minority in the usual sense,
rather a qualitative majority in the terminology of Nimzowitsch.

14Nbd5 15 Kb1 Qc7 16 Rhf1

The position is roughly balanced. Black has an extra pawn, but his task to
convert the pawn is more than difficult. If he just starts to exchange off
the bishops and knights, he might find himself on the defensive, with the
backward pawn on f7. On the other hand, Whites active pieces and the
pressure on pawn f7 allow him to play on with confidence and ambition.

5. Think for Yourself

Now and then we should develop a new idea by ourselves. Lev Gutman
didnt regard my idea as correct. Nevertheless, it looks interesting.
Nothing for your next correspondence tournament, perhaps, but we
shouldnt underestimate the stress factor in OTB chess, or overestimate
Blacks reflexes in a competition with fast time limits.

1 d4 d5 2 e4 dxe4 3 Nc3 Nf6 4 f3 c6! 5 Bc4 exf3 6 Nxf3 Bf5! 7 g4!??!

Since we are in the Blackmar-Diemer Gambit, sporadic excesses with

the ! and ? marks are permitted. As far as I can tell, the text move is new.
Please let me know, if you ever played it. Now if 7Bxg4?, White
replies 8 Bxf7+ Kxf7 9 Ne5+. And 7Bg6 8 h4 can hardly be critical
either.

7Nxg4 8 Nh4 g6

The most natural reaction:

(a) 8...e6 9 Nxf5 exf5 10 0-0 Nf6 11 Rxf5, and White is better.

(b) 8...Bd7 doesnt impress either. White may continue: 9 Ne4 b5 10 Qe2
e5 11 Bxf7+ Kxf7 12 Qf3+ Nf6 13 0-0 exd4 14 Qb3+ Ke7 15 Re1, etc.

(c) 8...Bc8!? is less natural than the text move, but also critical:

(c1) 9 0-0?! Nf6 10 Nf3

Should it really be worth a second pawn to deny the Bc8 an active role
and gain one tempo? An interesting idea, but unlikely, e.g., 10e6 11
Qe2 Nbd7 12 Bf4 Be7 13 Rae1 Nf8.

(c2) 9 Ne4 Nf6 (9Nd7 10 Qxg4 unclear; 9e5 10 0-0!) 10 Rf1 (10
Nxf6+ SWJediknight [6] 10exf6! 11 0-0 Be7) 10Nbd7 11 Qe2 e6
(11Nb6 12 Nxf6+ gxf6 13 Bxf7+ Kxf7 14 Qh5+ Ke6 =) 12 Bg5 Qa5+
(12Be7 13 Nf5 exf5 14 Nd6+ Kf8 15 Nxf7, about =) 13 Bd2 Qh5 14
Nxf6+ Nxf6 15 Nf3 Bd7 (15Nd5 16 Bxd5 Qxd5 17 Ne5 f6 18 c4 Qxd4
19 Rxf6!? Qh4+ 20 Rf2 Bc5 21 Be3 Bd6 22 Nxc6!? unclear) 16 Rg1 (my
original analysis went 16 0-0-0 0-0-0 17 Rde1 with compensation, but
SWJediknights 16Nd5! followed by f6 and an eventual Qh5-e8 [6] is
strong) 16h6 17 0-0-0 =+

White seems to have some compensation for the sacrificed pawns, e.g.,
17Nd5 18 Qf2 or 170-0-0 18 Rdf1 Be8 19 a3 Rg8 20 Qe1.

9 Ne4 Nf6 10 Rf1 Nbd7 11 Qe2 e6 12 Bg5 Qa5+ 13 Bd2 Qh5 14 Nxf6+
Nxf6 15 Nf3 Bd7 16 0-0-0 0-0-0 17 Rde1.

9 Bg5

Apparently more precise than 9 Nxf5 gxf5 10 Bg5 Qd6 or 10Rg8!,

MNb [6]. However, 10 h3 Nf6 (SWJediknight [6]) 11 Bg5 e6 12 d5 may
be worth a look.
9Qd6

There is no lack of alternatives:

(a) 9h6 10 Nxf5 gxf5 (10hxg5 11 Qxg4 gxf5 12 Bxf7+!?) 11 Bh4

Ne3 (11b5 12 Bb3 b4 13 Ne4!?; 11Bg7 12 Qe2) 12 Bxf7+ Kxf7 13.
Qh5+ and then

(a1) 13...Ke6 14 d5+ Kd7 15 Bf2 (15 Rd1 Nxd1 16 Qxf5+ Kc7 17 Qe5+
Kc8 18 Kxd1 Rh7 19 Qf5+ e6 20 Qxh7 Qxh4) 15...Nxc2+ (15...Nxd5 16
Bd4 Rg8 17 Nxd5 cxd5 18 Qf7 Rg5 19 h4 Rg2 20 0-0-0 Kc8 21 Bc5 e6
22 Bxf8 Qd7 23 Be7 Nc6 24 Qf8+), and here either 16 Ke2 Qe8 17 Qxf5
+ e6 or 16 Kd2 Nxa1 17 Qxf5+ Kc7 18 Qe5+ Qd6 19 Qxh8. White has
reasonable chances for a draw, although Black certainly has an advantage.

(a2) 13Kg8 14 Qg6+ Bg7 15 Qe6+ Kh7 16 Rg1 Rg8 17 Qxe3 Qxd4 18
Qxd4 Bxd4 19 Rxg8 Kxg8 20 0-0-0, about =.

(b) 9Bc8 10 Qf3 Nf6 11 Bxf6 exf6 12 0-0-0 Bh6+ 13 Kb1 0-0 14 Qg3
Kh8 15 d5 with sufficient compensation.

(c) 9Bg7 10 h3 h6 11 Bf4 Nf6 12 Nxf5 gxf5 13 Rg1 Rg8 14 Qd3, and
White has a lot of play for the two pawns.

(d) 9Qd7 10 Nxf5 gxf5 (10Qxf5 11 Qd2 f6 12 Bf4 g5 13 Bxb8 Qf3!

14 Rf1 Qe3+ 15 Qe2 Rxb8 16 Be6 Qxe2+ 17 Kxe2 h5 18 Rad1; for
example, 18Bg7 19 h3 Nh6 20 Ne4 Rd8 21 d5 with good chances for a
draw) 11 h3 Bh6 12 Qd2 Bxg5 13 Qxg5 Qxd4 14 hxg4 Qxc4 15 Rxh7
Qd4 16 Rxh8+ Qxh8 17 0-0-0 Nd7 18 gxf5 Qe5 19 Qg8+ Nf8 20 Qg5
Rd8 21 Rf1.

Blacks extra pawn can hardly be converted, e.g., 21a5 22 a3 b5 23

Qg2 with counterplay.

10 Qd2!

10 Nxf5 gxf5 11 Qd3 (11 Qd2? Bh6) 11e6 12 0-0-0!? Nf2 13 Qe3
Nxd1 14 Rxd1. However, this attack fails: 14Rg8! 15 d5 Be7 16 Bxe7
Qxe7 17 dxe6 fxe6 18 Bxe6 Qg5! +.

10b5

Perhaps cautious play is more appropriate: 10Rg8!? 11 0-0-0 Nd7 12

Kb1 e6 13 d5 cxd5 14 Bxd5 Nge5 15 Qe2 Qb4 or 10h5 11 0-0-0 Nd7
12 h3 Ngf6 13 Kb1 Be6 14 Bf4 Bh6 15 Bxh6 Bxc4 16 Bf4 Qb4 come
into consideration. In both cases Black seems to have a slight advantage.

11 Bb3 a5 12 Bf4 e5

(a) 12Qf6 13 Nxf5 gxf5 14 0-0-0 Bh6 15 a4 b4 16 Ne2 Bxf4 17 Nxf4.

(b) 12Qd8 13 a3 Rg8 14 Nxf5 gxf5 15 d5.

(c) 12Qd7 13 Nxf5 gxf5 14 a3 (after 14 a4?! b4 15 Ne2 Na6, Black

will castle long, when his king will find a safe place on b7) with unclear
complications.

13 h3 exf4

Or 13b4 14 hxg4 exf4 15 gxf5 bxc3 16 Qxc3 a4 17 Bxf7+ Kxf7 18

Nxg6 hxg6 19 Rxh8 Bg7 20 Rh7 Kg8 21 fxg6 Qxg6

22 Rxg7+ Kxg7 23 0-0-0 unclear.

14 hxg4 Bxg4

14Bd7!? 15 0-0-0! a4 16 Bxf7+ Kxf7 17 Ne4 Qc7 18 Nf3 Bg7 19 Ne5

+ Kg8 20 Qxf4 Be8 21 g5 Nd7 22 Nf6+ Bxf6 23 gxf6 Bf7 24 Qh6 Nxf6
25 Rdf1 =.

15 Ne4 Qe7

After 15Qc7 16 Nf6+ Kd8 17 Nxg4 a4, comes the surprising 18 Be6!
Bg7 (18fxe6 19 Nxg6 Rg8 20 Nxf4 +=) 19 Bf5 Qe7+ (19gxf5 20
Nxf5 Bf8 21 d5 +=) 20 Ne5 f6 21 0-0-0 fxe5 22 Bxg6 e4 23 Nf5! +=.
16 Qxf4 f5 17 Kf1 fxe4 18 Qxg4 Kd8 19 Ke2! Bh6 20 Raf1 Rf8 21 Ng2

21Rxf1 22 Rxf1 Qg5 23 Qxg5 Bxg5 24 Bg8, about =.

I cant deny that 7 g4 is risky. More analysis will be necessary to decide

whether the move is sound. Fortunately, White has a reliable and more
positional alternative in Lev Gutmans idea 7 Bg5 Be7 8 Nh4 Bg6 9 Nxg6
hxg6 10 Qd3!. And we shouldnt forget the equalizing 5 Nxe4.
Altogether, it seems that I had overestimated the OKelly Defense in my
article [4] and [5].

Sources:
[1] E. J. Diemer: Das moderne Blackmar-Diemer-Gambit, Band 1,
Heidelberg 1976.
[2] V. Hergert: Die OKelly-Verteidigung im Blackmar-Diemer-Gambit,
Dsseldorf 1993
[3] G. Studier: Emil Joseph Diemer, Dresden 1996.
[4] St. Bcker: Rettung aus grauem Elend durch Diemers Gambit, in:
Kaissiber 5 (1998)
[5] St. Bcker: Kritische Varianten im BDG, in: Kaissiber 8 (1998)
[6] Thread Antidotes to the Blackmar-Diemer Gambit (2009- ) [7]
Analyses by Emanuel Rajmund on the colorful reference site for the
Alchemy Variation.

Send your games or comments to redaktion@kaissiber.de.

[ChessCafe Home Page] [Book Review] [Columnists]

[Endgame Study] [The Skittles Room] [Archives]
[Links] [Online Bookstore] [About ChessCafe.com] [Contact Us]

2009 CyberCafes, LLC. All Rights Reserved.

"ChessCafe.com" is a registered trademark of Russell Enterprises, Inc.